With increasing popularity of automation and intelligent electronic devices, such as computerized machines, IoT (the Internet of Things), smart vehicles, smart phones, drones, mobile devices, airplanes, artificial intelligence (“AI”), the demand of intelligent machine and faster real-time response are increasing. For machine learning to become mainstream, a significant number of pieces, such as data management, model training, and data collection need to be improved.
A conventional type of machine learning is, in itself, an exploratory process which may involve trying different kinds of models, such as convolutional, RNN, attentional, et cetera. Machine learning or training typically concerns a wide variety of hyper-parameters that change the shape of the model and training characteristics. Model training generally requires intensive computation. As such, real-time response via machine learning model can be challenging.